Workforce development has far stronger causal evidence for investment than investing in entrepreneurship does.
Most conversations about economic investment start with entrepreneurship. New companies, new jobs, new disruption. It is the dominant frame in policy, philanthropy, and platform economics. Universities build incubators. Cities fund accelerators. Foundations back founder cohorts. The assumption underlying all of it is that backing entrepreneurs is how you grow an economy.
The evidence does not support that assumption. And the contrast with what the evidence does support is stark enough that it should change how serious investors think about economic development.
What the evidence actually shows
For the past four decades, economists have subjected workforce development programs to some of the most rigorous evaluation in social science. Randomized controlled trials — the same evidentiary standard used to approve pharmaceutical drugs — have measured the causal impact of training programs on earnings, employment, and economic mobility with precision that almost no other investment category can match.
The results are extraordinary. Year Up, a sectoral training program placing low-income young adults into corporate IT and financial services roles, produced a 30% increase in annual earnings sustained through year seven — the largest earnings gain ever documented in a high-quality workforce RCT, according to the federally sponsored PACE evaluation conducted by Abt Associates for the U.S. Department of Health and Human Services. Per Scholas, which trains unemployed adults for technology careers at a cost of approximately $8,000 per participant, reports an 8 to 1 economic return according to its own annual data, while an independent ten-year randomized trial found cumulative earnings gains of roughly $42,000 against that cost. Project Quest, a healthcare and skilled trades program in San Antonio, was found in a 14-year randomized controlled trial by the Economic Mobility Corporation to have generated $54,000 in cumulative earnings gains per participant against an average investment of $16,244 — a 234% return, with participants moving from poverty-level wages to nearly $60,000 annually by the final year of the study. The Hendren and Sprung-Keyser welfare analysis published in the Quarterly Journal of Economics, which evaluated 133 historical policy changes using a unified framework, found that the best sectoral training programs more than pay for themselves through increased tax revenue and reduced transfer payments.
Investing in Workforce Development is backed by the strongest causal evidence of any category of economic investment that I know of. Not the strongest in workforce policy. The strongest anywhere.
Now look at what we have for entrepreneurship incubators
The same Harvard framework that rigorously evaluated 133 government policy changes contains zero incubator or accelerator evaluations. Not because researchers tried and found no effect. Because the data required to measure them was never collected.
The Kauffman Foundation, the largest private foundation dedicated to entrepreneurship research in the United States, examined more than 35 academic articles on incubators and concluded that they might not prove more effective at creating successful businesses than unincubated businesses. A 2024 study by Wharton professors Valentina Assenova and Raphael Amit, published in the Strategic Management Journal, examined data from 8,580 startups across 408 accelerators in 176 countries and found that accelerated startups were just 3.4% more likely to raise follow-on venture capital than unaccelerated ones.
Meanwhile, more than 75% of venture-backed companies never return their investors’ capital, according to Harvard Business School senior lecturer Shikhar Ghosh’s research on more than 2,000 ventures. The top 10% of VC-backed companies generate approximately 90% of all returns, according to Commonfund’s analysis of 35 years of venture data. This power law concentration is known and accepted by venture capitalists — it is their operating model. But most institutional investors in entrepreneurship are not venture capitalists. They are universities, foundations, governments, and workforce programs funding cohorts as if each founder has roughly equal odds, with no portfolio logic and no rigorous outcome measurement.
The accountability asymmetry is where this becomes a policy problem. Workforce programs funded under the federal Workforce Innovation and Opportunity Act must report against six standardized performance indicators using unemployment insurance wage records and independent administrative data. On the other hand, incubator and accelerator programs receiving equivalent public funds face no comparable requirements. Metrics are self-reported. Comparison groups are absent. Survivorship bias is endemic — programs report successful exits while failed ventures disappear from the data entirely.
We hold workforce investment to some of the highest evidentiary standards in social science and give entrepreneurship investment a complete pass on sentiment alone.
This is not an argument against entrepreneurship
Joseph Schumpeter, the early twentieth century Austrian economist who gave us the concept of creative destruction, was right that entrepreneurship is the primary engine of long run economic growth. The evidence supports him at the macro level. And there are narrow conditions under which entrepreneurship investment is clearly the superior bet.
NYU Stern professor Sabrina Howell found, in a quasi-experimental study of the Department of Energy’s SBIR grant program published in the American Economic Review in 2017, that an early-stage grant approximately doubles a startup’s probability of attracting subsequent venture capital, with large positive effects on patenting and revenue. These effects were strongest for financially constrained firms — the mechanism was funding technology prototyping, not certifying quality. Elite accelerators with intensive mentoring components produce measurable value when the program is genuinely selective and the mentoring is substantive.
But Schumpeter’s own framework identifies the conditions under which disruption generates economic returns. The innovation has to be genuinely disruptive, not incremental. It has to create more than it destroys across the broader economy. And the economic cycle has to favor new disruption rather than the absorption of existing innovations. Most institutional investment in entrepreneurship fails all three tests. And there is no rigorous measurement infrastructure in place to know when it does and when it does not.
What this means for anyone allocating capital toward economic impact
Workforce development does not get the headlines. It does not produce unicorns. It does not generate the cultural narrative of the scrappy founder changing the world. But it reliably raises wages for people who need higher wages, generates returns that exceed its costs, and has been proven to do so across decades of independent evaluation.
The programs that work share a common structure: they train for specific industries with documented labor shortages, they secure employer commitments before training begins, they provide wrap-around support for participants who face economic barriers, and they track outcomes rigorously over time. Year Up. Per Scholas. Project Quest. These are not experimental models. They are proven ones.
The question for any institution — foundation, university, city government, platform company, ecosystem investor — that genuinely wants to produce economic impact is whether the investment they are making meets the evidentiary bar that workforce development has already cleared. Most entrepreneurship programs do not come close. If your goal is to make society measurably better, the evidence says invest in workforce development.
Postscript
Claude helped me form this article, but true credit should go to the Schumpeter chapter of The Worldly Philosophers by Robert L. Heilbroner. Heilbroner put Schumpeter into terms that made me instantly draw a dichotomy between VC and workforce investment. Claude then helped me build an app that used Schumpeterian calculus, which led to me finding very few instances where an investor should invest in Innovation versus iorkforce development.
